Martin Englund och Fredrik Thuring
presenterar sitt examensarbete

Multidimensional Experience Rating with Extensions to Causal Time Dependence

Abstract:
The subject of this master thesis is to use individual claim information to
improve claim prediction. This is done by generalizing the well known
Bühlmann-Straub credibility estimator to incorporate information from
additional policies of the individual. A time dependence is also introduced
to let claim prediction be influenced by the age of claim information. The
derived estimators are implemented and applied on a data set obtained from
the Swedish insurance company Trygg-Hansa's portfolio of private persons.
The main conclusions are that individual claim information can be used to
improve claim prediction with a credibility estimation approach and that
information from other policies of the individual does not contribute to
the claim prediction.